YIG Oscillators: Selection and Specification

A YIG-tuned oscillator (YTO) is a microwave source whose frequency is set by a magnetic field rather than by a voltage or a physical dimension. A YIG sphere sits in the field of an electromagnet, resonating at a frequency proportional to that field, and an active device supplies the negative resistance that sustains oscillation. Change the coil current and the output frequency follows — linearly, across an octave or more. That single property is why YTOs remain the default wideband source in test equipment and electronic warfare receivers decades after cheaper alternatives arrived.

Where it sits in the chain

A YTO is a frequency source, one of the families collected under microwave and RF components. In a receiver it is usually the first local oscillator, feeding a mixer or converter; in test equipment it is the sweeper. Its two defining characteristics — very wide tuning and low phase noise — are exactly what a broadband receiver front end needs, because the LO has to reach every frequency the receiver covers without polluting the conversion.

How it works, briefly

The full mechanism is covered in how YIG oscillators work and the material physics in what YIG is. In short:

• The YIG sphere is the frequency-determining resonator, with high unloaded Q from its narrow ferrimagnetic resonance linewidth.
• The active device supplies gain and negative resistance across the band; it does not set frequency.
• The main coil provides coarse, wide, linear tuning. It is inductive, therefore slow.
• The FM coil provides narrow, fast correction, and is what a phase-locked loop drives.

The practical consequence of that split is worth stating plainly, because it is where most integration effort goes: you are not buying an oscillator, you are buying a control system. The crystal is the part that behaves well. The coil, the driver, the thermal path and the calibration are the parts that need engineering.

Key specifications

Parameter	What it means	What it constrains
Tuning range	Frequency span reachable with the main coil	Whether one source covers your band, or you need several
Output power	Level at the connector, typically flat-ish across band	Mixer drive level and whether you need a buffer amplifier
Phase noise	L(f) at a stated offset and carrier	Receiver reciprocal mixing and achievable noise figure
Tuning linearity	Deviation of frequency from a straight line versus coil current	How much calibration and correction the system must carry
Hysteresis	Frequency difference approaching a point from above versus below	Absolute frequency accuracy without recalibration
Settling time	Time to reach and stay within a tolerance after a step	Maximum sweep or hop rate
Operating temperature	Range over which specifications hold	Thermal design and the correction table you need

Parameter meanings are general to the device class. For values, work from the specific part's datasheet — ranges vary widely across models and grades.

Across the commercial electromagnet YTO range, published figures give a sense of the envelope: tuning spanning roughly 700 MHz to 40 GHz across models, output levels around +9 to +17 dBm, phase noise on quieter models near −123 to −130 dBc/Hz at 100 kHz offset, and standard operating temperature of 0 to +65 °C with −20 to +70 °C and −40 to +85 °C available.

Treat those as the shape of the envelope, not as a specification for any part you might buy.

The two coils, and why the split matters when you specify

When a datasheet quotes a settling time, check which coil it refers to. A full-band main-coil step and a small FM-coil correction differ by orders of magnitude, and a vendor is entitled to quote whichever is flattering. If your system sweeps, the main coil figure is the one that bounds you. If it locks and holds, the FM coil figure is.

Reading the phase noise number honestly

The headline dBc/Hz figure is the most-quoted and least-useful number on a YTO datasheet, because it is quoted at whichever offset flatters the part.

What matters is the offset your system integrates over. A receiver rejecting a strong adjacent signal cares about the mid-range offsets where reciprocal mixing happens. A coherent processor cares close-in. A wideband downconverter cares about the far-out floor because that sets its noise figure. Understanding phase noise covers how to read the plot rather than the headline.

One consequence specific to YTOs: because they are fundamental oscillators rather than multiplied sources, they avoid the 20·log₁₀(N) penalty a multiplier chain pays. A YTO that looks unremarkable against a multiplied source’s headline number can still be the quieter option at frequency.

Against the alternatives

Criterion	YIG-tuned oscillator	Varactor-tuned VCO	Dielectric resonator oscillator
Tuning range	Octave or more	Wide, but typically less	Essentially fixed
Tuning linearity	Very linear in current	Non-linear in voltage	Not applicable
Phase noise	Low across a wide band	Higher, degrades with tuning range	Very low at its fixed frequency
Settling time	Milliseconds (main coil)	Sub-microsecond	Not applicable
Power consumption	Continuous coil current	Low	Low
Size	Large; magnet dominates	Small, often a single package	Small
Magnetic sensitivity	Significant	Negligible	Negligible
Typical use	Wideband LO, sweepers	Agile synthesis, PLL blocks	Fixed low-noise reference

The pattern is consistent: YIG buys range and purity, and pays in speed, power and size. If your frequency plan is fixed, a DRO is quieter for less money. If you must hop in microseconds, no amount of YIG engineering will get you there. The full decision, including a decidable conclusion, is YIG oscillator vs VCO vs DRO.

Application matrix

The same part is a good or bad choice depending entirely on what the system is doing. This is the matrix worth arguing about before anyone looks at a datasheet.

Application	What dominates the choice	Where a YTO struggles
Spectrum / signal analyser LO	Full-band coverage and sweep linearity	Sweep rate at the fastest span settings
EW and SIGINT receiver LO	Wideband coverage with low reciprocal mixing	Hop speed against agile threats
Radar exciter	Close-in phase noise and stability	Frequency agility between pulses
SATCOM up/downconverter LO	Phase noise at the loop offsets; long-term stability	Power budget and size in the chassis
Automated test sweeper	Tuning linearity and repeatability	Test time when settling dominates the sequence
Spectrum monitoring front end	Coverage plus a tracking preselector	Preselector and LO settling must be budgeted together

Qualitative. Every row is a judgement about which parameter binds first, not a vendor recommendation.

The last row is worth expanding, because it is the case people most often underestimate. A monitoring front end usually pairs a YTO with a YIG-tuned preselector, and both are magnetically tuned, so both settle slowly. Their settling times add into the same budget. A preselector holds a constant absolute bandwidth of about 20 to 40 MHz across octaves at 1 to 3 dB insertion loss, and that insertion loss lands directly on the system noise figure — so the preselector is buying you blocking performance at a measurable cost, and the LO has to be good enough to justify it.

Selection criteria

That last question is not an afterthought on a component with a long design life. See obsolete RF component sourcing for how to handle it.

Advantages and limitations

Advantages. Octave-plus tuning from one device. Highly linear frequency-versus-current. Low phase noise across the full range rather than at one point. Fundamental operation, so no multiplication penalty.

Limitations. Millisecond settling on wide steps. Continuous power dissipation to hold frequency. Physically large because the magnet dominates the package. Sensitive to external magnetic fields and to its own magnetic history. Requires a current driver, and usually calibration.

Typical applications

Wideband receivers and spectrum analysers, where the LO must reach everything the front end covers. Electronic warfare and signal intelligence receivers, for the same reason under harder constraints. Automated test equipment sweepers. Radar systems where a clean, widely tunable source outweighs hop speed — see radar RF components.

Buying and lifecycle

YTOs are low-volume, long-life parts, and that shapes procurement more than the specification does.

• Specify the assembly, not the resonator. Oscillator plus driver plus calibration is the unit that has a performance figure. Buying the pieces separately transfers the integration to you, along with the responsibility for the phase noise number.
• Ask what the datasheet figure was measured with. A phase noise plot is taken with a particular driver, at a particular carrier, at a particular temperature. Any of those changing changes the number.
• Check production status early. Programme lifetimes in this field routinely outrun component lifetimes, and a magnetically tuned source is not a drop-in replacement for another vendor’s magnetically tuned source. The calibration is part specific.
• Plan for the calibration data. Whatever corrects linearity and drift has to be stored, versioned and reproducible on a repaired unit. That is a systems problem, not a component one.

The general version of this discipline — traceability, date codes, incoming inspection, obsolescence — is in sourcing and quality.

Testing and integration notes

• Measure phase noise on the locked assembly you will ship, not the bare oscillator, if it runs in a loop.
• Characterise hysteresis by approaching test frequencies from both directions; the difference is the number your calibration has to absorb.
• Verify settling against your actual step size. A settling figure for a full-band step tells you little about a 10 MHz correction, and vice versa.
• Check frequency accuracy across the enclosure temperature range, not at ambient.
• Confirm magnetic clearance in the mechanical layout before the enclosure is fixed, not after.

Related guides

• What is YIG — the material and the tuning mechanism
• How YIG oscillators work — the loop, the coils and the driver
• YIG oscillator vs VCO vs DRO — when the alternatives win
• Understanding phase noise — how to read the parameter that usually decides the choice
• YIG-tuned filters — the usual tracking preselector
• Radar RF components — how the same source is constrained by a waveform

Turn the data sheet into an acceptance plan

A YIG oscillator data sheet is a starting condition, not a complete production test. Each quoted result is tied to a carrier frequency, output level, temperature, driver configuration and measurement bandwidth. Copy those conditions into the system requirement before comparing units. A broad tuning-range claim does not say where output power is lowest; a phase-noise curve does not say what the control driver adds; and a settling figure does not say which step, error band or approach direction was used.

Build acceptance around the jobs the source will perform. At several frequencies across the intended band, command a controlled approach and record frequency error, output level and lock state. Make representative small and large steps, then measure time to the receiver or transmitter’s actual error tolerance rather than to an unspecified “settled” indication. Check the same points at the enclosure temperature limits, because coil resistance and magnetic behaviour shift with temperature.

Include the tuning ports. The main coil should be checked for range and repeatability; the FM or correction coil should be checked for sufficient authority without adding unacceptable modulation noise. If the source is locked, test it with the real reference level and test reacquisition after a reference interruption. These are not optional accessories to the oscillator: they determine the output a system uses. The how YIG oscillators work guide details why the paths are split.

Retain the evidence with the serial number. Store the calibration table, driver revision, reference condition, test temperature and plots used to accept the assembly. This makes a later repair or alternate source manageable: the replacement can be tested against the same operating definition rather than against a generic catalogue line. When procurement risk is high, apply the traceability and incoming-inspection steps in obsolete RF component sourcing before the source reaches a difficult-to-service platform.